Quantitative Identification of Yellow Rust, Powdery Mildew and Fertilizer-Water Stress in Winter Wheat Using In-Situ Hyperspectral Data

摘要

As the progressive effects of global warming, the yield loss caused by diseases and pests are increasing in winter wheat. It is necessary to distinguish different diseases for guiding variable rate spraying in wheat. Nevertheless, it is very difficult to quantitatively identify different diseases and fertilizer-water stress by specific sensitive bands selected from multi spectral data over a large area. Conversely, hyper spectral data contain more information, and provide the potential for quantitative identification of different stresses. This study focused on identification and distinction of yellow rust, powdery mildew and fertilizer-water stress by canopy spectral reflectance. Fifteen commonly used vegetation indices were selected, and independent t-test was done to get sensitivity index for each stress. Finally, a combination index was optimally selected to distinguish the three stresses. The results showed that the integrative index (NDVI-PhRI) combining normalized difference vegetation index (NDVI) and physiological reflectance index (PhRI) could be used to identify powdery mildew and yellow rust (PM-YR). A 2-dimensional spatial coordinate was established based on the NDVI and PhRI derived from hyper spectral data, then the different stress data were displayed in the spatial coordinate and the classification boundary could be used to identify the powdery mildew and yellow rust stress. Similarly, yellow rust and fertilizer-water stress (YR-nOwO) can be distinguished by the combination index (MSR-PhRI) derived from modified simple ratio (MSR) and physiological reflectance index (PhRI); and the combination index (NRI-RVSI) derived from nitrogen reflectance index (NRI) and red-edge vegetation stress index (RVSI) was accurate to identify powdery mildew and fertilizer-water stress (PM-nOwO). For the PM-YR, YR-nOwO and PM-nOwO models, their verification accuracies were 83.3%, 88%, 88.75%, and the kappa accuracies were 63.41%, 74.79%, 71.43%, respectively. It indicated that the combination index derived from hyperspectral data could be used to identify the different stresses and provide guides for crop management across a large area.